考虑QoS属性相关性的Web服务选择

Web Service Selection Based on Correlation of QoS Attributes

现有的Web服务选择方法通常假定偏好由用户给出.由于偏好的主观性和模糊性,用户通常无法用具体数字表达清楚自己的偏好.且QoS各维属性之间存在相关性,偏好加权的方法无法消除信息的重叠,导致服务综合QoS评价不准确.针对该问题,在Web服务选择框架中对QoS属性设置区间搜索以考虑用户的优先偏好,使得初选的服务满足用户的QoS约束.对初选的服务利用主成分分析的思想,提出一种可行的Web服务选择算法PCA-WSS,根据各主成分的贡献率进行加权,分离QoS各维属性之间存在的相关性,有效地评价服务的综合QoS,为用户选择综合QoS最优的服务.实验结果验证算法的有效性和可行性.

The existing approach of Web service selection assume that preferences of users have been offered by users. However, con- sidering the subjectivity and fuzziness of preferences, users can hardly offer clear preferences. QoS attributes may be interrelated, so weighted summation approach cannot eliminate correlations among QoS attributes, which leads to inaccurate results. To solve the problem, a searching mechanism with QoS range setting is proposed in Web service selection framework to consider user＇s preference priority, thereby making primarily selected services satisfy the user＇s QoS constraints. With the primarily selected services, based on the idea of principal component analysis （PCA）, a feasible algorithm of Web service selection named PCA-WSS （Web Service Se- lection Based on PCA ） is proposed, which can eliminate correlations among QoS attributes, evaluate the comprehensive QoS of Web services more effectively and select the Web service with the best comprehensive QoS for users. Finally, the effectiveness and feasi- bility of,our approach are verified by experiments.